So I have a pandas data frame similar to this:
| col1 | col2 | col3 |
|---|---|---|
| [0,1,0] | 1 | 0 |
| [1,0,0] | 0 | 1 |
and I want to unpack it so that it becomes 5 columns as apparently tensorflow doesn't support nested lists. Is there an efficient way to achieve this?
-
Can you provide a formatted example of your data? and also provide an explanation to what you mean with "multiple Tensorflow inputs"AloneTogether– AloneTogether2022-04-20 05:50:49 +00:00Commented Apr 20, 2022 at 5:50
-
1@AloneTogether sorry about the formatting, didn't understand that it needed spaces but it's working now. For what I'm trying to achieve, I'm trying to convert col1 into something like col1-1, col1-2, col-3 etc. as 3 new columns, because TensorFlow doesn't really like there being columns of integers and lists at the same time.ChronoVortex– ChronoVortex2022-04-21 11:57:17 +00:00Commented Apr 21, 2022 at 11:57
-
@AloneTogether yes, thank you!ChronoVortex– ChronoVortex2022-04-23 04:52:10 +00:00Commented Apr 23, 2022 at 4:52
Add a comment
|
1 Answer
You can try merging the lists with pandas:
import pandas as pd
df = pd.DataFrame(data = {'col1': [[0,1,0], [1,0,0] ], 'col2': [1, 0], 'col3': [0, 1]})
df['col1-1'], df['col1-2'], df['col1-3'] = zip(*list(df['col1'].values))
df = df.drop('col1', axis=1)
print(df)
col2 col3 col1-1 col1-2 col1-3
0 1 0 0 1 0
1 0 1 1 0 0
Or with numpy:
import pandas as pd
import numpy as np
df = pd.DataFrame(data = {'col1': [[0,1,0], [1,0,0] ], 'col2': [1, 0], 'col3': [0, 1]})
col1 = np.vstack(df['col1'].values)
col23 = df[['col2', 'col3']].values
data = np.concatenate([col1, col23], axis=-1)
print(data)
[[0 1 0 1 0]
[1 0 0 0 1]]